• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of virLink to Publisher's site
J Gen Virol. Sep 2008; 89(Pt 9): 2063–2074.
PMCID: PMC2886952

Recombination, decreased host specificity and increased mobility may have driven the emergence of maize streak virus as an agricultural pathogen

Abstract

Maize streak virus (MSV; family Geminiviridae, genus Mastrevirus), the causal agent of maize streak disease, ranks amongst the most serious biological threats to food security in subSaharan Africa. Although five distinct MSV strains have been currently described, only one of these – MSV-A – causes severe disease in maize. Due primarily to their not being an obvious threat to agriculture, very little is known about the ‘grass-adapted’ MSV strains, MSV-B, -C, -D and -E. Since comparing the genetic diversities, geographical distributions and natural host ranges of MSV-A with the other MSV strains could provide valuable information on the epidemiology, evolution and emergence of MSV-A, we carried out a phylogeographical analysis of MSVs found in uncultivated indigenous African grasses. Amongst the 83 new MSV genomes presented here, we report the discovery of six new MSV strains (MSV-F to -K). The non-random recombination breakpoint distributions detectable with these and other available mastrevirus sequences partially mirror those seen in begomoviruses, implying that the forces shaping these breakpoint patterns have been largely conserved since the earliest geminivirus ancestors. We present evidence that the ancestor of all MSV-A variants was the recombinant progeny of ancestral MSV-B and MSV-G/-F variants. While it remains unknown whether recombination influenced the emergence of MSV-A in maize, our discovery that MSV-A variants may both move between and become established in different regions of Africa with greater ease, and infect more grass species than other MSV strains, goes some way towards explaining why MSV-A is such a successful maize pathogen.

INTRODUCTION

Maize streak virus (MSV; family Geminiviridae, genus Mastrevirus) is best known as the causal agent of maize streak disease. Although the geographical range of MSV is largely restricted to subSaharan Africa, its serious impact on food security in the world's poorest countries ranks it amongst the most important agricultural pathogens globally (Bosque-Perez, 2000).

Although a significant degree of MSV diversity has been documented (Dekker et al. 1988; Clarke et al., 1989; Pinner & Markham, 1990; Peterschmitt et al., 1991; Martin et al., 2001; Willment et al., 2001), only one low diversity strain, called MSV-A (McClean, 1947; Storey & McClean, 1930), is responsible for maize streak disease (Pinner et al., 1988; Briddon et al., 1994). MSV-A variants are generally the only viruses sampled from field-collected maize plants presenting with severe streak disease, and the only variants known to cause the disease under laboratory conditions (Martin et al., 2001).

Other strains of MSV – called MSV-B to -E, but often collectively referred to as ‘grass-infecting’ or non-maize-adapted MSVs – have only very rarely been isolated from maize plants, and generally only produce very mild symptoms in even the most MSV-sensitive maize genotypes (Pinner et al., 1988; Martin et al., 1999, 2001; Schnippenkoetter et al., 2001; Willment et al. 2002). While this might suggest that these other MSV strains are largely irrelevant to African agriculture, they have been found infecting other cultivated crops such as wheat, rye, barley and oats (Willment et al., 2001). Under laboratory testing conditions, MSV-B and -C are significantly more virulent than MSV-A isolates in wheat and barley (Schnippenkoetter et al., 2001; Willment et al., 2002), suggesting that these viruses may have a substantial, albeit an as yet undetermined, impact on African agriculture.

Besides their potential agricultural relevance, the non-maize-adapted MSV strains may hold important clues to the past and ongoing evolution and epidemiology of the maize-adapted MSV-A strain. Given that the most prevalent MSV-A variant in southern Africa – MSV-A4 – is actually a MSV-A/B recombinant (Martin et al., 2001), there is a clear precedent for the non-maize-adapted MSV strains directly contributing via recombination to ongoing MSV-A evolution. More thorough analyses of MSV recombination involving a greater diversity of MSV full-length genomes could, as has been done with other geminiviruses (Lefeuvre et al., 2007b; Prasanna & Rai, 2007), provide a more detailed picture of sequence exchange patterns most commonly associated with evolutionary advancement of MSVs. Also, from both an ecological and evolutionary perspective, comparative analyses of geographical and host range distributions of different MSV strains could help identify unique aspects of MSV-A epidemiology that have facilitated its emergence as an important agricultural pathogen.

We therefore undertook a survey of MSV diversity in indigenous uncultivated African grass species. Amongst 83 MSV isolates sampled in West Africa, East Africa, southern Africa and La Réunion, we identified six new MSV strains. We present evidence of extensive inter-strain MSV recombination and identify recombination breakpoint hot- and cold-spots that are partially conserved amongst all geminiviruses. Our analysis reveals significant differences in the natural host and geographical ranges of various MSV strains that may have a bearing on the emergence of MSV-A as a maize pathogen.

METHODS

Virus sampling.

Grasses displaying symptoms characteristic of MSV infection were sampled between 2005 and 2007 from South Africa (n=39), Zimbabwe (n=1), Mozambique (n=4), Nigeria (n=9), La Réunion (n=8) and Uganda (n=7). We also analysed archived samples collected in previous studies between 1986 and 2000 from Nigeria (n=3), Burundi (n=1) and Rwanda (n=1; Pinner et al., 1988; Pinner & Markham, 1990), Mali (n=1), Zimbabwe (n=2; Peterschmitt et al., 1991) and South Africa (n=7; Willment et al., 2001). Only one sample was taken per host species in any given sampling location and different sampling locations were separated by two or more kilometres (see Supplementary Table S1 available in JGV Online for sampling coordinates, dates and host species).

Cloning and sequencing of full genomes.

Viral genomes were isolated from plant material as described previously (Owor et al., 2007a; Shepherd et al., 2008a). Amplified concatemers were digested with either BamHI, KpnI or SalI to yield ~2.7 kb linearized viral genomes that were ligated into pGEMZf+ (Promega Biotech). Both strands of cloned genomes were commercially sequenced (Macrogen) using the primer set described by Owor et al. (2007a).

Host species identification.

Host species were identified by chloroplast ndhF gene sequencing. C-terminal encoding portions (~ 1.1 kb) of the ndhF genes were amplified from grass genomic DNA using the PCR primers: 972-F (5′-GTCTCAATTGGGTTATATGATG-3′) and 2110-R (5′-CCCCCTAYATATTTGATACCTTCTCC-3′) using Kapa Taq (Kapa Biosystems) described by Olmstead & Sweere (1994) and Giussani et al. (2001). The ndhF amplicons were ligated into pGEM-T Easy (Promega Biotech) and completely sequenced by Macrogen using M13 standard sequencing primers.

Sequence analyses.

All available African streak virus genome sequences, including those of Urochloa streak virus (Oluwafemi et al., 2008), Eragrostis streak virus (Shepherd et al., 2008b), Panicum streak virus (Briddon et al., 1992; Varsani et al., 2008), Sugar cane streak virus (Hughes et al., 1993; Shepherd et al., 2008b), Sugar cane streak Egypt virus (Bigarré et al., 1999) and Sugar cane streak Reunion virus (Bigarré et al., 1999; Shepherd et al., 2008b), were obtained from public sequence databases. Sequence alignments were constructed using the poa program (Grasso & Lee, 2004) and edited both by eye and using the clustal w-based (Thompson et al., 1994) sequence alignment tool implemented in mega (version 4; Tamura et al., 2007). mega was also used to calculate the pairwise sequence identities shared by aligned genomes using pairwise deletion of gaps.

Maximum-likelihood phylogenetic trees were constructed using the phyml program (Guindon & Gascuel, 2003). The F81+G4 nucleotide substitution model was selected as being the most appropriate for the analysis of MSV evolution using the modeltest web server (Posada, 2006)

Recombination was analysed using the rdp (Martin & Rybicki, 2000), geneconv (Padidam et al., 1999), bootscan (Martin et al., 2005a), maxchi (Smith, 1992), chimaera (Posada & Crandall, 2001), siscan (Gibbs et al., 2000) and 3seq (Boni et al., 2007) methods implemented in the rdp3 program (Martin et al., 2005b). Default settings were used throughout and only potential recombination events detected by two or more of the above methods, coupled with phylogenetic evidence of recombination, were considered significant. The severity of Bonferroni correction was minimized by only searching for recombination signals in a single sequence within groups of sequences sharing >99.3 % sequence identity. Using the approach outlined in the rdp3 program manual (http://darwin.uvigo.es/rdp/rdp.html), the approximate breakpoint positions and recombinant sequence(s) inferred for every potential recombination event were manually checked and adjusted where necessary using the phylogenetic and recombination signal analysis features available in rdp3.

The distribution of unambiguously detected breakpoint positions of all unique recombination events was analysed for evidence of recombination hot- and cold-spots with rdp3 as described by Heath et al. (2006). Published rdp3 project files describing breakpoint distributions detectable in bipartite and monopartite begomoviruses (Lefeuvre et al., 2007b) were merged in rdp3 and used to produce a composite plot of begomovirus recombination breakpoint distributions.

RESULTS AND DISCUSSION

Discovery of new MSV strains

We cloned and fully sequenced 83 individual MSV genomes sampled primarily from indigenous African grasses presenting with streak symptoms. For preliminary objective classification of these sequences we aligned them with a selection of MSV (n=88) and non-MSV (n=24) African streak virus genomes and determined pair-wise percentage sequence identities shared between them. All of the new sequences shared greater than 79.1 % identity with previously described MSV isolates (Supplementary Table S2 available in JGV Online) and, based on the current ICTV species demarcation guidelines for the mastreviruses (Stanley et al., 2005), they are all MSV strain isolates.

Consistent with previous analyses of African streak virus diversity (Martin et al., 2001; Willment et al., 2002; Shepherd et al., 2008b; Varsani et al., 2008), we further subdivided the sequences into strain groupings. As is clearly indicated by a deep trough between 92 and 94 % identity in a plot of pair-wise MSV sequence identities (Supplementary Fig. S1 available in JGV Online) and, in accordance with Martin et al. (2001), we identified 93 % identity as a ‘natural’ MSV strain demarcation threshold. Using this criterion we classified the MSV isolates into 11 strains (named MSV-A to -K), only five of which (MSV-A to -E) have been described previously (Martin et al., 2001; Schnippenkoetter et al., 2001; Willment et al., 2002).

The similarities between the newly determined sequences and previously described MSV isolates allowed us to deduce that they contained all genomic features that have previously been identified as having functional relevance during MSV infections.

Evidence of extensive inter-strain MSV recombination

As recombination features prominently in geminivirus evolution (Lefeuvre et al., 2007b; Martin et al., 2001; Padidam et al., 1999; Prasanna & Rai, 2007) and can cause phylogeny reconstruction errors (Awadalla, 2003; Penny et al., 2007; Posada & Crandall 2002) we tried to remove, as far as possible, the influence of recombination from the construction of an MSV phylogeny. We therefore analysed the 83 newly sequenced genomes together with all other 112 publicly available African streak virus genomes using a battery of seven recombination analysis methods implemented in the rdp3 program. We found clear evidence of 36 distinct recombination events (detectable by three or more different analysis methods and with good phylogenetic support) spread across 164 of the 195 analysed genomes. Twenty-seven of these events were detected in 157 of the 172 analysed MSV sequences (Fig. 1; Supplementary Table S3 and Supplementary rdp3 project file).

Fig. 1.
Complex relationships amongst MSV isolates sampled throughout Africa and the Indian Ocean island of La Réunion. Whereas tree branches with less than 50 % bootstrap support have been collapsed, those with greater than 70 and 90 % ...

Due to such a high proportion of the analysed MSV sequences being detectably recombinant (91.3 %), it proved more difficult to produce a recombination-free MSV phylogeny than we had anticipated. The primary problem was that there was no fraction of the sequence alignment longer than 519 nt (alignment positions corresponding to nt sites 311 and 760 in MSV-Ns, taking position 1 as the first A residue 3′ of the virion strand origin of replication) that was unbroken by detectable recombination breakpoints in any of the sequences. As this 519 nt region is quite conserved amongst the MSV isolates, phylogenetic analyses focusing on it lacked sufficient power to resolve relationships amongst individuals within particular strains (see Supplementary Fig. S2 for this recombination-free phylogeny).

To provide a general description of the relationships between all of the sequences we therefore opted to simply construct a maximum-likelihood tree using the full genome sequences, essentially ignoring recombination, and present it together with a breakdown of the sequence mosaics that must be considered when interpreting its topology (Fig. 1).

Despite recombination undermining our confidence in the accuracy of this MSV phylogeny, there is good bootstrap support (>70 %) for all of our tentative MSV strain classifications. It should, however, be pointed out that the recombination analysis indicated that only four of the 11 strains (MSV-B, MSV-E, MSV-G and MSV-I) are predominantly represented by sequences that are not the products of inter-strain recombination events involving exchanges of more than 30 % of their genomes.

Some of the recombinant strains, such as MSV-H and -F, appear to have quite complex mosaic structures. For example, Ng-Lag-2007, the only MSV-H isolate we have sampled, has a genome that appears to have been assembled during at least four separate recombination events. Adding to the complexity of interpreting the origins of sequences such as Ng-Lag-2007 is that we cannot know, without better sampling, either how old many of these recombination events are, or the order in which they most likely occurred. For example, lack of clear evidence for where the different pieces of Ng-Lag-2007 have come from indicates that the detected recombination events occurred between either progenitors of the sampled strains (i.e. if they are older events), or divergent, currently unsampled MSV genotypes (i.e. if they are more recent events).

Possibly the most interesting amongst the less complex recombinant strains is MSV-A – the strain that causes maize streak disease. Previous analyses of recombination amongst MSV strains (Martin et al., 2001; Padidam et al., 1999) failed to detect that this strain had arisen from an ancient recombination event between MSV-G/MSV-F and MSV-B progenitors. This was because without the newly discovered MSV-G and -F genomes it was not possible to tell that the MSV-A virion sense ORFs were unusually similar to those of the MSV-Bs. Every currently sampled MSV-A genome has an unmistakable trace of this recombination signal (including the most divergent genomes from La Réunion; P=7.4×10−9), indicating that the recombination event must have occurred prior to the time of the last common ancestor of all known MSV-As.

We must, however, caution that, given the inherent difficulties associated with identifying recombinant sequences in datasets with such high degrees of recombination, it is possible that we have misidentified MSV-A as the recombinant in this sequence exchange. That a recombination event has occurred is very probable, but we cannot be absolutely certain that it is not all of either the MSV-B or MSV-G and MSV-F sequences that are recombinant instead of the MSV-A sequences. It may require either the discovery of non-recombinant close relatives of the MSV-A viruses or fitness studies on laboratory reconstructions of the possible ancestral parental and recombinant viruses to prove that it is MSV-A and not the other strains that are recombinant. However, until this information becomes available, the most parsimonious hypothesis presented by RDP3 is that the MSV-As are recombinant.

Partial conservation of recombination patterns amongst geminiviruses

Conserved patterns of inter-species geminivirus recombination including recombination hot- and cold-spots have recently been described amongst members of the genus Begomovirus (Lefeuvre et al., 2007a, b; Prasanna & Rai, 2007). Although similarities in intra-strain recombination rates have been demonstrated for MSV-A and variants of the begomovirus species, East African cassava mosaic virus and East African cassava mosaic Kenya virus (Owor et al., 2007b), it is currently unknown whether patterns of inter-species/strain recombination are also conserved between begomoviruses and mastreviruses. We therefore analysed the distribution of breakpoints detected in our African streak virus dataset using the method described by Heath et al. (2006). Despite the relatively low number of unique recombination events analysed (36 in the African streak virus dataset compared with 284 collectively detected in the datasets analysed by Lefeuvre et al., 2007b) there was strong statistical evidence of recombination hot-spots near the coat protein gene (cp)/short intergenic region (SIR) interface and at the virion-strand origin of replication (v-ori), and a recombination cold-spot spanning almost the entire cp (Fig. 2a). Importantly, these hot- and cold-spots are in almost precisely the same locations as those detected previously in begomoviruses (Fig. 2b). Unlike with the begomoviruses, however, in the mastrevirus genomes there is no evidence of a clear recombination hot-spot near the centre of the replication associated protein gene (rep). Also, in the mastrevirus dataset the cp/SIR interface hot-spot is substantially more pronounced than the v-ori hot-spot, whereas the converse is true for the begomoviruses.

Fig. 2.
Partial conservation of recombination breakpoint distributions across the family Geminiviridae. (a) Breakpoint distribution plot (solid black line) indicating recombination hot- and cold-spots detectable in African streak ...

Despite these differences, this result strongly suggests that similar processes are shaping recombination breakpoint distributions in both genera. Importantly, the observed recombination patterns, including the differences between them, are entirely consistent with recent hypotheses that have invoked a mixture of biochemical and selective forces to explain non-random recombination breakpoint distributions in geminiviruses (Jeske et al., 2001; Lefeuvre et al., 2007a, b).

In both the begomoviruses and mastreviruses the recombination hot-spots map to complementary-sense gene transcription initiation and termination sites and virion-strand origins of replication. The reason complementary gene transcription initiation and termination sites may be more predisposed to recombination than other sites is possibly that these are the regions where the most frequent clashes between transcription and replication complexes occur (Lefeuvre et al., 2007a). The absence in mastreviruses of a transcription initiation site and promoter elements in the middle of rep analogous to those found in begomoviruses (Shung et al., 2006) may explain why there is no detectable recombination hot-spot in this region of mastrevirus genomes.

It is also possible that the absence of a short intergenic region in begomoviruses could be the reason for the recombination hot-spot mapping to the 3′ end of cp in begomoviruses being smaller than that detected in mastreviruses. The distribution of recombination breakpoints detectable in our mastrevirus dataset is consistent with proposals that breakpoint distributions observed in geminiviruses sampled from nature are strongly influenced by selective forces that eliminate recombinants with defective intra-genome interactions (Martin et al., 2005c; Lefeuvre et al., 2007b). It has been convincingly demonstrated that selection strongly favours the survival of geminivirus recombinants in which both intra-protein amino acid interactions (Lefeuvre et al., 2007b) and inter-genome region interactions (Martin et al., 2005c) remain undisrupted. Importantly, there are various lines of evidence that indicate that recombination breakpoints both at the 3′ end of cp (García-Andrés et al., 2007a; Lefeuvre et al., 2007b) and within the SIR (Martin et al., 2005c; Martin & Rybicki, 2002) are particularly undisruptive of intra-genome interactions. MSV SIR sequences are extremely modular and can continue functioning properly even when transferred into genetic backgrounds very different from those in which they evolved (Martin & Rybicki, 2002). If the modestly sized recombination hot-spot at the 3′ end of the begomovirus cp is caused by the coincident location of a site encoding a proportion of CP that tolerates recombination well at the same position as a transcription terminator that is biochemically predisposed to recombination (due to clashes between transcription and replication complexes), then it is reasonable to suspect that the placement of a highly modular intergenic region beside this site in mastreviruses is responsible for the larger size of this hot-spot in these viruses.

Differences in the geographical distributions of MSV strains and variants

We were interested in determining whether there were differences in MSV strain demographics in different parts of Africa analogous to those previously detected for cassava-infecting geminiviruses (Ndunguru et al., 2005; Bull et al., 2006). We therefore split the sampled viruses into southern African (isolates from Zimbabwe, South Africa and Mozambique; n=70), West African (isolates from Nigeria and Mali; n=11), East African (isolates from Uganda, Rwanda, Burundi and Kenya; n=10) and La Réunion (n=8) groups and tested for differences in the strain compositions of these groups. Although there was a highly significant difference in the strain distributions across all four regions collectively (P=8.3×10−8, 4 [regions]×11 [MSV strains] χ2 test), separate pairwise comparisons between the regions indicated that these differences originated primarily from the West African population sample (Fig. 3). While the East African and southern African strain distributions were also significantly different, the West African strain distribution is clearly the most unique. The key differences between the sampled West African MSV population and those found elsewhere are the absence of any MSV-B isolates and the presence of MSV-G and -H isolates.

Fig. 3.
MSV strain demographics in different parts of Africa and La Réunion. All virus isolates represented here were sampled from uncultivated grass species. Different strains are represented by different colours, and P values indicate regions with significantly ...

This was a surprising result as we had anticipated that the mainland African MSV-populations would have similar structures, and that the La Réunion population would be distinct. Recent reports of large numbers of new geminivirus species, strains and variants unique to the Indian Ocean islands (Delatte et al., 2005; Lefeuvre et al., 2007a; Peterschmitt et al., 1996; Shepherd et al., 2008b) have indicated that they have been reasonably isolated with respect to the movement of geminiviruses. Conversely, the close relationships shared by MSV strain A isolates found in West Africa with those found elsewhere on the continent (Briddon et al. 1994; Martin et al., 2001) clearly indicate that the movement of at least some MSV strains either to or from West Africa is relatively frequent and largely unhindered. Our failure to sample any MSV-B isolates in West Africa and any MSV-G and -H isolates outside West Africa therefore suggests that there may be strain-specific differences in the continent-wide movement of MSV variants.

To investigate this possibility further we compared the phylogenies of MSV-A and -B isolates (the two best sampled MSV strains) in the context of their regions of origin (Fig. 4). Both MSV-A and -B isolates from particular regions tend to group in phylogenetic trees with other isolates from the same region. However, all MSV-B isolates from particular regions form monophyletic groups (i.e. all viruses from, for example, southern Africa are more closely related to other southern African viruses than they are to East African or Réunion viruses), which is not the case for the MSV-A isolates. MSV-A isolates sampled in different parts of Africa are polyphyletic in that, for example, different groups of southern African isolates are more closely related to West and East African isolates than they are to other groups of southern African isolates. This difference between the MSV-A and -B phylogenies strongly suggests that, over the evolutionary timescales represented by these trees, MSV-A variants are moving between and becoming established in different parts of Africa at a greater rate than MSV-B variants. This implies that there are ecological barriers to the movement of MSV-B variants across Africa that are not experienced by MSV-A variants and provides good support for our suggestion that there are strain-specific differences in the continent-wide movement of MSV variants across Africa.

Fig. 4.
Phylogenetic evidence that MSV-A variants move between and become established in different regions of Africa more frequently than MSV-B variants. (a) Relationships amongst MSV-B sequences from wild and cultivated grass species in East ...

Given that the MSV-A, -B and -C strains all share common vector species with other African streak virus species (Schnippenkoetter et al., 2001; Willment et al., 2002; Bigarré et al., 1999; Briddon et al., 1992) it would be reasonable to suppose that all of the other newly discovered strains also probably share the same vector species. It might therefore seem obvious that strain specific differences in host ranges and/or degrees of host adaptation are probably responsible for differences in their geographical distributions. It cannot, however, be ruled out that the different MSV strains are specifically adapted to transmission by different vector races or biotypes (Mesfin et al., 1991) and that differences in the feeding preferences and geographical distributions of these races and biotypes might underlie differences in MSV strain distributions.

Also, while we have dealt here with MSV isolates sampled from uncultivated grasses, it is important to point out that the distributions of cultivated MSV host species such as maize, wheat, sugar cane and millet might also have an important impact on the continental spread of different MSV strains. For example, a major host of MSV-A is maize and the widespread distribution of this cultivated species has possibly aided the movement of MSV-A throughout the continent. Another possibility that should be considered with crop-infecting viruses such as MSV is that they might be transported directly by humans. MSV-A isolates have recently been identified infecting sugar cane throughout large parts of South Africa (van Antwerpen et al., 2008). As infections are characteristically mild and sugar cane is vegetatively propagated, it is possible that inadvertent transportation of infected stalks might accelerate the movement of MSV-A variants. It is even conceivable that transportation of MSV-A-infected maize cobs might facilitate the movement of this strain. Although MSV is not seed transmitted, immature maize cobs are frequently transported within their leaf-like sheaths. These ‘green envelopes’ display streak symptoms in MSV-infected plants and the virus could therefore presumably be acquired by leafhoppers should they feed on them. However, good phylogenetic evidence of at least some MSV-A diversification along geographical lines indicates that if long-distance human transportation of MSV-A across the continent occurs at all, it is probably infrequent.

Host range variation amongst MSV strains

In an attempt to directly determine whether differential host preferences (either by virus strains or vector biotypes) might at least partially account for differences in the geographical distributions of different strains, we analysed the strains in the context of the hosts from which they were isolated. Grouping hosts by genus (11 groups excluding cultivated host species) and viruses by strain (10 groups – excluding MSV-D for which no clear host identification could be made) we found very strong statistical support (P<1×10−8; 11×10 χ2 test) for significant differences between the hosts from which members of different MSV-strains were sampled.

However, we noted that we had oversampled certain host genera (e.g. Digitaria species, n=39) and undersampled others (e.g. Axonopus species, Rottboellia species and Pennisetum species, n=1 each; Supplementary Fig. S3 available in JGV Online). We also realized that on the island of La Réunion we had apparently oversampled Cenchrus species relative to other regions (3/4 of all Cenchrus samples were obtained on this island). Importantly, we found evidence of significant differences between the host types sampled in different regions [P=0.027, 10 (host genera)×4 (geographical region) χ2 test]. This sampling bias was primarily accounted for by the greater numbers of Cenchrus species sampled from La Réunion as it could be resolved by removing the La Réunion sample from the analysis (P=0.149). Given that only MSV-B isolates were sampled on La Réunion and that the sampling bias was only marginally significant, we did not anticipate that it would have a substantial effect on our assessment of the prevalence of different MSV strains in different host species. It is important to point out, however, that our sampling of obviously symptomatic plants may have unpredictably biased our analysis of natural host range distributions, in that unsampled host species/MSV strain combinations associated with mild or asymptomtic infections might be just as epidemiologically relevant as those associated with severely symptomatic infections.

We sought to offset some of these potential sampling biases by investigating the frequencies with which different MSV strains were sampled only in the three host genera for which we obtained seven or more samples: Digitaria species (n=39), Urochloa species (n=8) and Setaria species (n=7; Fig. 5). We found significant differences in the MSV strains isolated from the plants of these three genera [P<1×10−8, 3 (host genera)×9 (virus strains−no MSV-D and -J isolates were sampled from the three genera considered) χ2 test]. Considering the host genera in pairs we only found a significant difference between the relative frequencies with which different MSV strains were sampled from Digitaria and Setaria plants (P<1×10−8). Whereas Setaria plants tended to be infected with the closely related MSV-K and -C strains, Digitaria plants tended to be infected with the closely related MSV-A, -B and -G strains.

Fig. 5.
The frequencies with which different MSV strains were sampled from three common groups of MSV hosts. Different strains are represented by different colours and P values indicate significant differences (values in bold) between the three ...

We then looked for specific differences between the apparent host ranges of different virus strains. Despite our sampling too few viruses to achieve enough statistical power to differentiate between the host ranges of most of the MSV strains, we observed significant natural host range differences between MSV-B and both MSV-C and -K [P=1.0×10−4 and 2.0×10−4, respectively; 2 (virus strains)×11 (host genera) χ2 test] and between MSV-G and both MSV-C and -K (P=2.9×10−2 and 5.0×10−2, respectively). Whereas MSV-B and -G isolates tended to come primarily from Digitaria species, MSV-K and -C isolates tended to come primarily from Setaria species. Should the MSV-C, -D and -K strains (Fig. 1) be separated into another species following the next revision of geminivirus taxonomic criteria, it seems logical, therefore, that the species be named Setaria streak virus and that these strains be renamed as SetSV-A, -B and -C, respectively (Fig. 1).

Although there was no significant difference in the range of host genera from which MSV-A and -B isolates were sampled (P=0.28), we noted that the 14 MSV-A samples were obtained from grasses in eight genera, whereas the 34 MSV-B samples were obtained from grasses in only six genera. While this may indicate that MSV-A has a broader host range than MSV-B – something that may explain differences in the continent-wide distribution of the strains – more intensive sampling in a greater variety of hosts will be required to conclusively prove this. Nevertheless, our discovery of MSV-A isolates infecting a variety of uncultivated grasses is significant in that these species are probably both the natural hosts from which this strain originally emerged as a maize pathogen, and represent the hosts that currently sustain it between maize growing seasons.

Conclusions

While our analysis of MSV diversity in uncultivated indigenous African grasses has revealed six new MSV strains, we have also detected for the first time, a degree of recombination amongst mastreviruses paralleling that seen in begomoviruses. Our demonstration that recombination patterns are partially conserved across the family Geminiviridae is particularly significant in that it indicates that early ancestral geminivirus genomes had largely the same recombinational predispositions and constraints as those experienced by modern geminiviruses. Of potentially greater immediate relevance, however, is our discovery that the maize-adapted MSV-A strain is possibly the product of an ancestral recombination event between Digitaria-adapted MSV-G/-F and MSV-B viruses. While this recombination event may have produced a virus with increased severity in maize – a host to which MSV-A seems particularly well adapted – our data are also consistent with the possibility that it may have enabled MSV-A to spread more efficiently throughout the continent by allowing it to infect a wider variety of hosts. Importantly, these hypotheses can be directly tested by reconstructing and analysing the virulence and host ranges of the ancestral MSV-A and its parental viruses. Despite the amount of speculation about how recombination may produce new viral species or strains with altered host ranges, cell tropisms or pathogenicities, there are actually very few well supported examples of this having occurred in nature (see Fondong et al., 2000; Pita et al., 2001; Monci et al., 2002; García-Andrés et al., 2007a, b for good exceptions). The possibility that a recombination event is ultimately responsible for the existence of maize streak disease certainly deserves thorough investigation as MSV-A might be an important example of how recombinational acquisition of novel traits can sometimes trigger pathogen emergence.

Supplementary Material

[Supplementary Material]

Acknowledgments

This research was funded by the National Research Foundation (South Africa). A. V. is supported by the Carnegie Corporation of New York. D. N. S. is supported by PANNAR (Pty) Ltd; D. P. M. is supported by the Wellcome trust; B. O. is supported by the Rockefeller foundation through the USHEPiA programme; A. L. M. is supported by the Canon Collins Trust for Southern Africa and a University of Cape Town International Scholarship. P. L. is supported by the French Ministère de la Recherche et de l'Enseignement Supérieur. J. M. L. is supported by CIRAD.

Footnotes

Published online ahead of print on 27 June 2008 as DOI 10.1099/vir.0.2008/003590-0.

The GenBank/EMBL/DDBJ accession numbers for the 83 MSV genome sequences are EU152254 and EU628564-EU628644.

Supplementary material is available with the online version of this paper.

References

  • Awadalla, P. (2003). The evolutionary genomics of pathogen recombination. Nat Rev Genet 4, 50–60. [PubMed]
  • Bigarré, L., Salah, M., Granier, M., Frutos, R., Thouvenel, J. C. & Peterschmitt, M. (1999). Nucleotide sequence evidence for three distinct sugarcane streak mastreviruses. Arch Virol 144, 2331–2344. [PubMed]
  • Boni, M. F., Posada, D. & Feldman, M. W. (2007). An exact nonparametric method for inferring mosaic structure in sequence triplets. Genetics 176, 1035–1047. [PMC free article] [PubMed]
  • Bosque-Perez, N. A. (2000). Eight decades of maize streak virus research. Virus Res 71, 107–121. [PubMed]
  • Briddon, R. W., Lunness, P., Chamberlin, L. C., Pinner, M. S., Brundish, H. & Markham, P. G. (1992). The nucleotide sequence of an infectious insect-transmissible clone of the geminivirus Panicum streak virus. J Gen Virol 73, 1041–1047. [PubMed]
  • Briddon, R. W., Lunness, P., Chamberlin, L. C. & Markham, P. G. (1994). Analysis of the genetic variability of maize streak virus. Virus Genes 9, 93–100. [PubMed]
  • Bull, S. E., Briddon, R. W., Sserubombwe, W. S., Ngugi, K., Markham, P. G. & Stanley, J. (2006). Genetic diversity and phylogeography of cassava mosaic viruses in Kenya. J Gen Virol 87, 3053–3065. [PubMed]
  • Clarke, B. A., Rybicki, E. P., Hughes, F. L., Kirby, R. & von Wechmar, M. B. (1989). Characterization of southern African isolates of Maize streak virus – typing of 3 isolates by restriction mapping. Intervirology 30, 86–95. [PubMed]
  • Dekker, E. L., Pinner, M. S., Markham, P. G. & van Regenmortel, M. H. V. (1988). Characterization of maize streak virus isolates from different plant species by polyclonal and monoclonal antibodies. J Gen Virol 69, 983–990.
  • Delatte, H., Martin, D. P., Naze, F., Goldbach, R., Reynaud, B., Peterschmitt, M. & Lett, J. M. (2005). South West Indian Ocean islands tomato begomovirus populations represent a new major monopartite begomovirus group. J Gen Virol 86, 1533–1542. [PubMed]
  • Fondong, V. N., Pita, J. S., Rey, M. E., de Kochko, A., Beachy, R. N. & Fauquet, C. M. (2000). Evidence of synergism between African cassava mosaic virus and a new double-recombinant geminivirus infecting cassava in Cameroon. J Gen Virol 81, 287–297. [PubMed]
  • García-Andrés, S., Tomas, D. M., Sanchez-Campos, S., Navas-Castillo, J. & Moriones, E. (2007a). Frequent occurrence of recombinants in mixed infections of tomato yellow leaf curl disease-associated begomoviruses. Virology 365, 210–219. [PubMed]
  • García-Andrés, S., Accotto, G. P., Navas-Castillo, J. & Moriones, E. (2007b). Founder effect, plant host, and recombination shape the emergent population of begomoviruses that cause the tomato yellow leaf curl disease in the Mediterranean basin. Virology 359, 302–312. [PubMed]
  • Gibbs, M. J., Armstrong, J. S. & Gibbs, A. J. (2000). Sister-Scanning: a Monte Carlo procedure for assessing signals in recombinant sequences. Bioinformatics 16, 573–582. [PubMed]
  • Giussani, L. M., Cota-Sanchez, J. H., Zuloaga, F. O. & Kellogg, E. A. (2001). A molecular phylogeny of the grass subfamily Panicoideae (Poaceae) shows multiple origins of C4 photosynthesis. Am J Bot 88, 1993–2012. [PubMed]
  • Grasso, C. & Lee, C. (2004). Combining partial order alignment and progressive multiple sequence alignment increases alignment speed and scalability to very large alignment problems. Bioinformatics 20, 1546–1556. [PubMed]
  • Guindon, S. & Gascuel, O. (2003). A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52, 696–704. [PubMed]
  • Heath, L., van der Walt, E., Varsani, A. & Martin, D. P. (2006). Recombination patterns in aphthoviruses mirror those found in other picornaviruses. J Virol 80, 11827–11832. [PMC free article] [PubMed]
  • Hughes, F. L., Rybicki, E. P. & Kirby, R. (1993). Complete nucleotide sequence of sugarcane streak Monogeminivirus. Arch Virol 132, 171–182. [PubMed]
  • Jeske, H., Lutgemeier, M. & Preiss, W. (2001). DNA forms indicate rolling circle and recombination-dependent replication of Abutilon mosaic virus. EMBO J 20, 6158–6167. [PMC free article] [PubMed]
  • Lefeuvre, P., Martin, D. P., Hoareau, M., Naze, F., Delatte, H., Thierry, M., Varsani, A., Becker, N., Reynaud, B. & Lett, J. M. (2007a). Begomovirus ‘melting pot’ in the south-west Indian Ocean islands: molecular diversity and evolution through recombination. J Gen Virol 88, 3458–3468. [PubMed]
  • Lefeuvre, P., Lett, J. M., Reynaud, B. & Martin, D. P. (2007b). Avoidance of protein fold disruption in natural virus recombinants. PLoS Pathog 3, e181. [PMC free article] [PubMed]
  • Martin, D. & Rybicki, E. (2000). RDP: detection of recombination amongst aligned sequences. Bioinformatics 16, 562–563. [PubMed]
  • Martin, D. P. & Rybicki, E. P. (2002). Investigation of Maize streak virus pathogenicity determinants using chimaeric genomes. Virology 300, 180–188. [PubMed]
  • Martin, D. P., Willment, J. A. & Rybicki, E. P. (1999). Evaluation of maize streak virus pathogenicity in differentially resistant Zea mays genotypes. Phytopathology 89, 695–700. [PubMed]
  • Martin, D. P., Willment, J. A., Billharz, R., Velders, R., Odhiambo, B., Njuguna, J., James, D. & Rybicki, E. P. (2001). Sequence diversity and virulence in Zea mays of Maize streak virus isolates. Virology 288, 247–255. [PubMed]
  • Martin, D. P., Posada, D., Crandall, K. A. & Williamson, C. (2005a). A modified bootscan algorithm for automated identification of recombinant sequences and recombination breakpoints. AIDS Res Hum Retroviruses 21, 98–102. [PubMed]
  • Martin, D. P., Williamson, C. & Posada, D. (2005b). RDP2: recombination detection and analysis from sequence alignments. Bioinformatics 21, 260–262. [PubMed]
  • Martin, D. P., van der Walt, E., Posada, D. & Rybicki, E. P. (2005c). The evolutionary value of recombination is constrained by genome modularity. PLoS Genet 1, e51. [PMC free article] [PubMed]
  • McClean, A. P. D. (1947). Some Forms of Streak Virus Occurring in Maize, Sugar-Cane and Wild Grasses. Pretoria: South Africa Government Printers.
  • Mesfin, T., den Hollander, J. & Markham, P. G. (1991). Cicadulina species and maize streak virus in Ethiopia. Trop Pest Manage 37, 240–244.
  • Monci, F., Sánchez-Campos, S., Navas-Castillo, J. & Moriones, E. (2002). A natural recombinant between the geminiviruses Tomato yellow leaf curl Sardinia virus and Tomato yellow leaf curl virus exhibits a novel pathogenic phenotype and is becoming prevalent in Spanish populations. Virology 303, 317–326. [PubMed]
  • Ndunguru, J., Legg, J. P., Aveling, T. A., Thompson, G. & Fauquet, C. M. (2005). Molecular biodiversity of cassava begomoviruses in Tanzania: evolution of cassava geminiviruses in Africa and evidence for East Africa being a center of diversity of cassava geminiviruses. Virol J 2, 21. [PMC free article] [PubMed]
  • Olmstead, R. G. & Sweere, J. A. (1994). Combining data in phylogenetic systematics: an empirical approach using three molecular data sets in the Solanaceae. Syst Biol 43, 467–481.
  • Oluwafemi, S., Varsani, A., Monjane, A. L., Shepherd, D. N., Owor, B. E., Rybicki, E. P. & Martin, D. P. (2008). A new African streak virus species from Nigeria. Arch Virol 153, 1407–1410. [PubMed]
  • Owor, B. E., Shepherd, D. N., Taylor, N. J., Edema, R., Monjane, A. L., Thomson, J. A., Martin, D. P. & Varsani, A. (2007a). Successful application of FTA classic card technology and use of bacteriophage phi29 DNA polymerase for large-scale field sampling and cloning of complete maize streak virus genomes. J Virol Methods 140, 100–105. [PubMed]
  • Owor, B. E., Martin, D. P., Shepherd, D. N., Edema, R., Monjane, A. L., Rybicki, E. P., Thomson, J. A. & Varsani, A. (2007b). Genetic analysis of maize streak virus isolates from Uganda reveals widespread distribution of a recombinant variant. J Gen Virol 88, 3154–3165. [PubMed]
  • Padidam, M., Sawyer, S. & Fauquet, C. M. (1999). Possible emergence of new geminiviruses by frequent recombination. Virology 265, 218–225. [PubMed]
  • Penny, D., White, W. T., Hendy, M. D. & Phillips, M. J. (2007). A bias in ML estimates of branch lengths in the presence of multiple signals. Mol Biol Evol 25, 239–242. [PubMed]
  • Peterschmitt, M., Reynaud, B., Sommermeyer, G. & Baudin, P. (1991). Characterization of maize streak virus isolates using monoclonal and polyclonal antibodies and by transmission to a few hosts. Plant Dis 75, 27–32.
  • Peterschmitt, M., Granier, M., Frutos, R. & Reynaud, B. (1996). Infectivity and complete nucleotide sequence of the genome of a genetically distinct strain of maize streak virus from Réunion Island. Arch Virol 141, 1637–1650. [PubMed]
  • Pinner, M. S. & Markham, P. G. (1990). Serotyping and strain identification of maize streak virus isolates. J Gen Virol 71, 1635–1640. [PubMed]
  • Pinner, M. S., Markham, P. G., Markham, R. H. & Dekker, E. L. (1988). Characterisation of maize streak virus – description of strains – symptoms. Plant Pathol 37, 74–87.
  • Pita, J. S., Fondong, V. N., Sangaré, A., Otim-Nape, G. W., Ogwal, S. & Fauquet, C. M. (2001). Recombination, pseudorecombination and synergism of geminiviruses are determinant keys to the epidemic of severe cassava mosaic disease in Uganda. J Gen Virol 82, 655–665. [PubMed]
  • Posada, D. (2006). ModelTest Server: a web-based tool for the statistical selection of models of nucleotide substitution online. Nucleic Acids Res 34, W700–W703. [PMC free article] [PubMed]
  • Posada, D. & Crandall, K. A. (2001). Evaluation of methods for detecting recombination from DNA sequences: computer simulations. Proc Natl Acad Sci U S A 98, 13757–13762. [PMC free article] [PubMed]
  • Posada, D. & Crandall, K. A. (2002). The effect of recombination on the accuracy of phylogeny estimation. J Mol Evol 54, 396–402. [PubMed]
  • Prasanna, H. C. & Rai, M. (2007). Detection and frequency of recombination in tomato-infecting begomoviruses of South and Southeast Asia. Virol J 4, 111. [PMC free article] [PubMed]
  • Schnippenkoetter, W. H., Martin, D. P., Hughes, F., Fyvie, M., Willment, J. A., James, D., von Wechmar, B. & Rybicki, E. P. (2001). The biological and genomic characterisation of three mastreviruses. Arch Virol 146, 1075–1088. [PubMed]
  • Shepherd, D. N., Martin, D. P., Lefeuvre, P., Monjane, A. L., Owor, B. E., Rybicki, E. P. & Varsani, A. (2008a). A protocol for the rapid isolation of full geminivirus genomes from dried plant tissue. J Virol Methods . [PubMed]
  • Shepherd, D. N., Varsani, A., Windram, O., Lefeuvre, P., Monjane, A. L., Owor, B. & Martin, D. P. (2008b). Novel sugarcane streak and sugarcane streak Réunion mastrevirus from southern Africa and La Réunion. Arch Virol 153, 605–609. [PubMed]
  • Shung, C. Y., Sunter, J., Sirasanagandla, S. S. & Sunter, G. (2006). Distinct viral sequence elements are necessary for expression of Tomato golden mosaic virus complementary sense transcripts that direct AL2 and AL3 gene expression. Mol Plant Microbe Interact 19, 1394–1405. [PubMed]
  • Smith, J. M. (1992). Analyzing the mosaic structure of genes. J Mol Evol 34, 126–129. [PubMed]
  • Stanley, J., Bisaro, D. M., Briddon, R. W., Brown, J. K., Fauquet, C. M., Harrison, B. D., Rybicki, E. P. & Stenger, D. C. (2005). Geminiviridae. In Virus Taxonomy, Eighth Report of the International Committee on the Taxonomy of Viruses, pp. 301–306. Edited by C. M. Fauquet, M. A. Mayo, J. Maniloff, U. Desselberger & L. A. Ball. London: Elsevier.
  • Storey, H. H. & McClean, A. P. D. (1930). The transmission of streak disease between maize, sugarcane and wild grasses. Ann Appl Biol 17, 691–719.
  • Tamura, K., Dudley, J., Nei, M. & Kumar, S. (2007). mega4: molecular evolutionary genetics analysis (mega) software version 4.0. Mol Biol Evol 24, 1596–1599. [PubMed]
  • Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994). clustal w: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673–4680. [PMC free article] [PubMed]
  • van Antwerpen, T., McFarlane, S. S., Buchanan, G. F., Shepherd, D. N., Martin, D. P., Rybicki, E. P. & Varsani, A. (2008). First report of Maize streak virus field infection of sugarcane in South Africa. Plant Dis 92, 982.
  • Varsani, A., Oluwafemi, S., Shepherd, D. N., Monjane, A. L., Owor, B., Windram, O., Rybicki, E. P., Lefeuvre, P. & Martin, D. P. (2008). Panicum streak virus diversity is similar to that observed for Maize streak virus. Arch Virol 153, 601–604. [PubMed]
  • Willment, J. A., Martin, D. P. & Rybicki, E. P. (2001). Analysis of the diversity of African streak mastreviruses using PCR-generated RFLPs and partial sequence data. J Virol Methods 93, 75–87. [PubMed]
  • Willment, J. A., Martin, D. P., Van der Walt, E. & Rybicki, E. P. (2002). Biological and genomic sequence characterization of Maize streak virus isolates from wheat. Phytopathology 92, 81–86. [PubMed]

Articles from The Journal of General Virology are provided here courtesy of Society for General Microbiology

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...